1) Field of the Invention
The present invention relates to a device and method for fingerprint identification, and a computer product that are capable of being installed in portable terminals such as a PDA (Personal Digital Assistant), a cellular phone, a PHS (Personal Handyphone System) terminal, a compact notebook-type personal computer, and the like.
2) Description of the Related Art
Conventionally, fingerprint identification devices are used to identify a person using his/her fingerprint as an image that is read by an image sensor. More specifically, at first, an image sensor reads a fingerprint of a person to be identified, and then an image of the fingerprint read is registered as a fingerprint image.
To actually identify a fingerprint, the image sensor reads the fingerprint of the person to be identified, and a matching degree is checked between the fingerprint image and a fingerprint image previously registered. If the matching degree is equal to or more than a threshold value, then the identity of the person is authenticated.
A pattern matching method is often used to calculate the matching degree between the two fingerprint images. The pattern matching method is generally used in the field of image processing. For example, assuming that I(x, y) represents a read fingerprint image of a person to be identified, and T(x, y) represents a registered fingerprint image previously registered, an evaluation value v(x, y) is obtained by the following expression.v(x, y)=(1/A)ΣΣI(i, j)T(i−x, j−y)  (1)where A is the overlapping area between I(i, j) and T(i−x, j−y). Coordinates (x, y) where v is the maximum indicate a position where the read fingerprint image and the registered fingerprint image match each other to the maximum extent, and the value at that time represents the matching degree.
If the read fingerprint image and the registered fingerprint image are binary images of white and black, I(x, y) and T(x, y) take the value of 0 or 1. The evaluation value at this time may be obtained by the expression (1) given above, or by the following expression (2). As explained above, the expression used to perform a correlation operation is not limited thereto, but has various forms.v(x, y)=(1/A)ΣΣnot(I(i, j) xor T(i-x, j-y))  (2)where “xor” represents an exclusive OR, and “not” represents negative, and A is the overlapping area between I(i, j) and T(i-x, j-y).
However, the pattern matching method used in the conventional fingerprint identification device has one significant problem. This is the overlapping area between the read fingerprint image and the registered fingerprint image.
In the expression (1) and the expression (2), a factor A corresponds to the area of the overlap. To calculate an accurate matching degree, the area needs to be divided by the factor A to normalize it. However, if the area of the overlap is small, the factor A as a denominator reduces, which causes the value of matching degree to be inaccurate. In order to obtain an accurate value, the area of the overlap is required to be about a half of the fingerprint image. Therefore, to obtain the larger area of the overlap, it is effective to use an image sensor having large area. Basically, by using an image sensor larger than the area of a finger, the overlap occupies most part of the image, and the problem does not occur.
However, recently, requirements for installing the fingerprint identification device in portable terminals such as PDA, cellular phone, PHS, compact notebook-type personal computer, and the like are significantly increasing.
This is because the functions of the portable terminal have become extremely advanced, and the information stored therein has increased, thereby increasing the risk of losing the information stored.
However, in these portable terminals, it is difficult to install an image sensor whose size is larger than a finger (larger than 10 mm square) rather than restriction to installation space. Particularly, in cellular phones and PDAs, an image sensor having the area of about a fraction of a finger can be installed. In the image sensor having such small area, it is extremely difficult to exhibit sufficient performance, and therefore, it is impossible to perform fingerprint identification with high accuracy.
A conventional art has been disclosed, for example, in Japanese Patent Application Laid-Open No. H10-187982.